Re: GNAT 3.15p transition plan
Florian Weimer <firstname.lastname@example.org> writes:
> I'm going to recompile the existing Ada packages in Debian using GNAT
> 3.15p, at least on x86.
> This involves the following steps:
> - Packaging GNAT 3.15p (mostly done). I'm going to omit DSO support
> (see below).
> - Fixing the FTBFS errors of current Ada packages (not necessarily
> 3.15p-specific). Drop DSO support, too.
> - Let someone upload the packages (I'm not a DD).
> - After some time, request removal of the existing and now unneeded
> DSO packages.
> Why no DSO support? The GNAT ABI changes in *each* release, and the
> run-time library can only be built with the corresponding GNAT
> version. This means that if we want true backwards compatibility
> (including a buildable libgnat-3.14p), we'd have to keep around GNAT
> 3.14p (and GNAT 3.15p and so on), and all libraries depending on it.
> This doesn't look acceptable to me.
> GNAT supports source-based (Ada) package repositories, but I think
> it's still worthwhile to add precompiled object code libraries, to cut
> down compilation time. The source repository approach is interesting
> once we want to support multiple GNAT installations in parallel (à la
> common-lisp-controller), though.
Florian, I don't agree. I understand your arguments but honestly,
there is no more than one ACT release per year. So people don't
have to often rebuilt their programs and libraries. Of course,
there aren't a lot of Ada applications in Debian but DSO is a
real benefit in terms of binary size. It would be a regression
to stop DSO support.
> What about other architectures besides x86? Honestly, I don't know.
> The only non-x86 Debian port I looked at was the SPARC one, and it
> is incomplete (it includes some Solaris-specific code and fails to
> build GNAT 3.x as a result). If there is demand for Ada on this
> architecture and someone shows me a proof that GNAT 3.14p is in
> reasonably good shape (e.g. good ACATS results), I'll try to make
> 3.15p competitive in this regard.
I think you should post this message to debian-devel in order
to get a larger audience, because so far, GNAT was not related
to GCC releases.